The choice to take cocaine can be altered by manipulating the amount and responses required to obtain drug and non-drug alternatives, and this research has contributed to development of behavioral treatments for drug abuse, such as contingency management. Contingency management is the most effective psychosocial treatment for substance abuse. However, the feasibility of contingency management could be enhanced by reducing its cost. Using a nonhuman model of behavioral therapy, we can thoroughly examine variables such as response requirement and reinforcer amount to determine arrangements that most effectively reduce drug choice. Effects of such variables on drug choice in nonhumans have been studied primarily with qualitatively similar response-reinforcer relations for both options. However, the choice to take a drug when drug and competing non-drug alternatives are available under qualitatively different circumstances represents a more translational approach to what human drug abusers experience. Behavior is more persistent with unpredictable (e.g., variable) relative to predictable (e.g., fixed) schedules When choice is between fixed and variable schedules of food, nonhumans tend to choose the variable option, even when the result is net reinforcer loss. Whether nonhumans will choose a variable vs. fixed schedule of drug delivery has not been examined. The overall goal of the proposed research is to examine whether unpredictable reinforcement is a determinant of cocaine choice in rhesus monkeys. The significance of this research is two-fold: 1) it will establish the extent to which a fundamental behavioral phenomenon-preference for unpredictable options-is a key feature of cocaine self-administration; 2) confirmation of our hypotheses would support the use of novel and easily implemented modifications to behavioral therapy for drug abuse based on the unpredictable nature of drug and non-drug options. Specific Aim 1 is to examine choice between predictable and unpredictable schedules of cocaine. It is hypothesized that cocaine will be a more effective reinforcer when available in an unpredictable situation. Specific Aim 2 is to examine choice between predictable and unpredictable schedules of cocaine and food delivery in a drug vs. food situation. It is hypothesized that cocaine will be more effective when it is available under an unpredictable schedule and less effective when food is available under and unpredictable schedule. Specific Aim 3 is to examine choice between cocaine and food when food is unpredictable in terms of schedule and amount per delivery. It is hypothesized that food will be more effective when its schedule and amount of delivery are unpredictable relative to situations where only schedule or amount is unpredictable. If cocaine is chosen more robustly under unpredictable situations, then the unpredictability of cocaine may serve to increase its reinforcing value. Conversely, reducing cocaine choice when alternative non-drug reinforcers are unpredictable may provide an effective and novel approach to treating cocaine abuse.